While artificial intelligence (AI) -based systems are being increasingly integrated into organizational decision-making processes, recent literature has introduced human-AI ensembles as a novel human-AI decision-making approach. In a human-AI ensemble, both humans and AI-based systems perform the same tasks of the decision-making process, with their decisions being aggregated into one final decision. However, despite the potential of human-AI ensembles to improve the quality of decisions, the potential side effects of ensembling human decisions with decisions from AI-based systems remain unclear. Drawing on identity literature, we investigate how ensembling human-AI decisions at work affects the human worker. We conducted an online experiment with 48 participants and found that ensembling human-AI decisions lowers the human’s competence-based self-esteem and thus decreases their satisfaction with the decision-making process. Furthermore, we found that the human’s perception of the difficulty of making the decisions moderates this effect in that a higher level of perceived difficulty leads to a higher loss of competence-based self-esteem. These findings provide valuable implications for managers and researchers who seek to optimize decision-making.
